MICHAEL A. S. GUTH, Ph.D., J.D. Managing Director, Risk Management Consulting send e-mail (E-mail is quickest method of contact).   116 Oklahoma Ave.   Oak Ridge, TN   37830-8604   Phone: (865) 483-8309

Financial Economics Homepage      ||       Attorney at Law Homepage

---

Are the Electricity Markets Becoming Better Indicators of Future Spot Market Prices? (Alternatively, Why Were the Futures Markets for Power So Dumb?)

© Copyright 2004 by Michael A. S. Guth. All Rights Reserved. No portion of this site, including the contents of this web page may be copied, retransmitted, reposted, duplicated, or otherwise used without the express written permission of Dr. Michael Guth. Reprinted from The Risk Desk (December 2001) with permission of the publisher, Scudder Publishing Group, LLC. www.scudderpublishing.com.

Historically, the forward and futures markets for power have given poor indication of future spot market prices; these futures markets seemed to get it all wrong about 70-80% of the time. Unless market participants were willing to pay outrageous premiums to lock in a price, the forward/futures markets for power offered little opportunity to unload price risk. Other commodities have had more efficient (read: better predicting) forward prices. By the end of 2001, there is good news on the horizon: the power futures markets are finally moving in line with expected spot market prices. Before we turn to the good news, we should see why the new alignment comes as welcome relief to past pricing gaffes in the futures markets.

Example 1. In February 2001, the Entergy July-August 2001 futures price hovered between $140-$150/MWh. The price was $130/MWh on April 27, 2001. Three business days later, the price fell to $112.50/MWh. By the end of May 2001, the price was down to $90/MWh. The actual spot prices in July and August 2001 averaged $32.29/MWh. Imagine how electric utilities and power marketing firms felt when they purchased blocks of power in February, March, and April of this year, only to see the value of these assets collapse a month or two later.

The precipitous fall in the futures price may be evidence of a speculative bubble. Back in 1998, I published a game theoretic model that depicted speculative bubbles in electricity markets.1 The game was played by separate groups of informed and uninformed trading firms, with the informed group having inside information about a unit outage or some other factor that could cause tight regional supplies for power. The uninformed group ended up buying whenever the informed bought, and as soon as the informed group realized their purchases induced this feedback effect, the seeds for a speculative bubble had been sown.

As another sign of a speculative bubble on the Entergy July-August 2001 futures contract, the August futures prices remained inflated (above $70/MWh) in the wake of July spot prices in the range of $20-$55/MWh range. Finally, we might conclude that the Entergy summer forward contract exhibited a speculative bubble if knowledge of fundamentals, e.g., weather forecasting, failed to explain the sharp drop in July-Aug power prices during the first weeks of May 2001. A North Carolina State University professor of atmospheric sciences informed Progress Energy that it is possible to obtain more accurate 10-day weather forecasts over those produced by the National Weather Service and similar organizations. However, no credible weather information for July and August would have been discernible as early as the first week of May.

Instead, the following sequence of events seems to have triggered the collapse in the July-Aug. futures price. The eastern United States experienced an unusually hot May this year. Power marketers watched to see if May spot prices would spike up above $150/MWh. When May prices failed to spike, the marketers realized that power supplies were plentiful. They concluded that supplies would remain plentiful in July and August, and that triggered the excess sales that brought July-August futures prices tumbling down in May.

Example 2. Let’s examine the trajectory of futures prices for the Cinergy April 2000 contract. On October 11, 1999, the futures price was $24.50/MWh. It moved down to $23.50/MWh on Nov. 30, 1999, and dropped as low as $22.80/MWh on Dec. 28, 1999. On January 31, 2000, the April futures price was $22.90/MWh. The actual average spot price in April 2000 was $27.72/MWh. The early futures market had a pronounced downward trend that turned out to underestimate the future spot market price.

Example 3. The Cinergy July-Aug 2002 futures price today is approximately $50/MWh. The two-year historical average for July and August combined is about $36/MWh. Cambridge Energy Research Associates forecasts that next summer’s Cinergy on-peak price will be $35/MWh. Does the market collectively know more than the forecasters and analysts, or is the futures market inefficient to peg the price at $50/MWh?

These gaps between the futures prices and the actual spot market average price remind us of the conditions that existed back in 1997. In 1997, the NYMEX and CBOE futures contracts traded at hubs in the eastern U.S. were just being established. Four years ago, power market participants expected a certain amount of price overshooting behavior and excessively high summer prices until the futures markets became more calibrated to regional supply and demand. It is surprising that despite the rapid evolution of power markets over the past four years, the efficiency of the power futures markets in 2001 is little better than it was in 1997. As the French say, “Plus ça change, plus c’est la même chose” (The more things change, the more they remain the same).

But for a number of contracts, the forward markets for power have accurately moved in line with expected spot prices. The Cinergy July-Aug forward price for 2002 is $50/MWh, compared to $47.50 for July-Aug of 2003. The lower price for the summer block in 2003 is consistent with new generating capacity lowering market-clearing prices. Indeed, the forward markets for power may capture trends (dynamics) better price levels (static analysis). And the summer forward market this year seems more efficient than it was last year at this time.

Arbitrage With a Storable Commodity. Some readers will look at these examples and say “So what?” They will argue that forward prices have no connection to future spot market prices. But that assertion contradicts basic principles of economics and flies in the face of empirical observation. We can illustrate the connection most clearly for a storable commodity. If the futures price becomes much larger than the current spot price, then the typical “cash-and-carry” arbitrage will set into the market. This arbitrage amounts to selling the futures, buying the spot, placing the remaining cash in the bank to collect interest, and carrying over until delivery date of the forward. Eventually, the arbitrage will drive the premium on the futures price to less than the carrying costs minus interest.

For a storable commodity, if the futures price dips far below today’s spot price, then a slightly more complicated strategy will set in. The strategy consists of short selling the spot, buying the cheaper futures contract, and using delivery of the commodity at the expiration of the futures contract to cancel the short spot market sale. This arbitrage strategy is more complicated, because the short sale might be called back before the trader can take delivery (next season) of the commodity purchased by futures contract.

Nevertheless, carrying costs minus interest define an upper bound for futures price deviations from spot prices. It remains an open empirical question as to whether there exists a symmetrical lower bound on futures price deviations below the spot price, due to the more complicated nature of the arbitrage that can reverse that pricing anomaly. The upper and hypothetical lower bound on futures prices for a storable commodity is illustrated in the following figure.

	Illustration of Forward Price Boundaries for a Storable Commodity

Keynesian Risk Premium. The futures price can remain within the upper and lower bounds, yet profitable trading can be executed. When the futures price exceeds the boundary conditions for a storability commodity, a riskless profit can be secured. However, traders frequently assume the risk of a long or short futures position as part of a trading strategy. Back in the 1920s and 1930s, the celebrated economist John Maynard Keynes made millions of British pounds by taking calculated risk (speculative) positions in the agricultural futures markets. Keynes hypothesized that agricultural commodity prices in the futures markets in which he speculated were consistently downward-biased estimators of the future spot market price.

Keynes argued that the lower futures price amounted to a risk premium required of the long hedgers (farmers with commodities to harvest) to transfer the price risk to the futures contract holders. In exchange for locking in a certain price for their future harvests, the farmers accepted a slightly lower price (on the order of a 5-10% risk premium discount) in the futures market than what they expected to receive if they took their chance on the future spot market. Keynes referred to this downward-biased spread between the futures price and the expected spot price as “normal backwardation.”

Arbitrage in Power Markets. Although power is not a storable commodity, riskless arbitrage is still possible in power markets. The key to riskless arbitrage is not storability but rather the ability to deliver on a contract at some future date or at some other location. Future delivery of power can be locked in, e.g., with a bilateral forward contract or with a tolling agreement on a generating plant. We witness arbitrage in the power markets nearly everyday, particularly in the hourly (spot) market. Academics and Wall Street financial experts with no knowledge of the power industry sometimes jump to the conclusion that the nonstorability of power means that most of modern asset pricing theory, which is underpinned by arbitrage arguments, is inapplicable to power.2 But modern asset pricing theory needs to be tailored to power markets, not discarded entirely.

In power markets, futures prices are tied (loosely) to expected spot prices. The Keynesian concepts of backwardation (downward-biased estimator) and in contango (upward-biased estimator) apply to power markets. Power marketers have incentives to obtain tolling agreements and other forward sales contracts when futures prices get too high. But tolling agreements and bilateral power sales are less flexible than “cash-and-carry” strategies, so the connection between power futures and spot prices might get blurred in some months.

In an ideal world, the futures price for power would deviate from the expected spot price by no more than 10% or 15%. Then the futures market would be an efficient means of predicting future excess demand or supply of power, and those who wished to unload price risk could obtain “price insurance” from the futures market at a reasonable premium. But it has taken several years for us to achieve reasonable efficiency in the power futures markets.

Caution With Futures Prices. Those who relied too much on the accuracy of power futures prices in 2001 suffered financial losses or embarrassment. The California state government and investment bank Salomon Smith Barney both relied heavily on power futures markets as an indicator of future spot prices. In May 2001, the California state government finally reversed its course and decided to purchase long-term power supply contracts for California consumers. Through 38 contracts totaling $43 billion, the state bought power at prices up to $154/MWh for the summer on-peak and $95/MWh for off-peak. The contracts varied widely in cost and complexity. Around June 11, 2001, hourly peak spot prices in California were $58/MWh.3

It should come as no surprise that the forward markets for power in California tumbled shortly after the state executed its purchases. Market analysts suggested that power suppliers knew about the quantities that the state intended to purchase from media reports and engaged in gaming behavior to jack up the price of power sold to the state. Attempting to prove or disprove these allegations will keep lawyers busy for years. But we know the power forward markets were very efficient at collecting large sums of California taxpayer funds; these markets were inefficient at reflecting underlying demand and supply.

California might have been caught up in hype over rising power prices fueled by reports such as “The Power Curve” by Salomon Smith Barney.4 Recall that the Entergy July-Aug 2001 futures contract traded earlier this year at $140/MWh, but the July-Aug spot prices eventually averaged $32.29/MWh. Back in September 2000, a Salomon report said “Despite the apparent onslaught of new generation capacity, forward electricity prices are higher than 2000 prices in nearly every region. Why is the forward price curve shifting higher? In our view, this is the result of unexpectedly high demand and tight supply.”

It turned out that the onslaught of new generation capacity contributed to excess supply and lower prices this year. The Salomon equity analysts were not wrong to believe there should be a connection between the futures price and the expected spot market price for power. They were wrong to believe that the power futures markets were already efficient at predicting future spot prices as early as September 2000.

This article noted at the outset that good news was on the horizon. The power forward markets seem to be getting closer to achieving a nexus between futures prices and expected spot prices. If 2002 is not radically different from the past few years in terms of native load and supply, then the historical average of volume-weighted spot market prices might be a reasonable first estimate for the expected spot market prices to prevail next year. In the period January 2002 – October 2002, the following Cinergy futures/forward prices today are within 15% of the four- or five-year historical averages.

Cinergy On-Peak	[Bid, Offer] Futures price ($/MWh) as of 12/17/2001	Historical Average Daily Price ($/MWh) for defined years
Jan-Feb 2002	[26.00, 26.10]	26.29 1998-2001
April 2002	[24.95, 25.25]	28.05 1997-2001
May 2002	[27.35, 27.75]	31.41 1997-2001
June 2002	[37.00, 37.25]	35.75 2000-2001
Sept. 2002	[25.10, 25.50]	23.04 1997-2000
Oct. 2002	[25.00, 25.45]	24.45 1997-2000
Nov. 2002	[25.00, 25.45]	26.62 1997-2000

In eight out of 12 months, the futures price appears reasonably close to the long term equilibrium and expected spot price. For 2002, we are interested in researching the percentage of trades at a given hub that are settled financially versus those trades taken to physical delivery. The PJM-West hub has a large percentage of transactions settled financially. It would be interesting to compare the futures prices at PJM-West with the futures prices at a hub where most of the transactions go to physical delivery. Intuitively, the hub with many financial transactions will probably be more efficient at predicting future spot prices than other hubs with fewer transactions and predominantly physical exchanges of power.